- Email: [email protected]
Analysis of alternating current driven electroluminescence in organic light emitting diodes: A comparative study
Organic Electronics 15 (2014) 1815–1821
Contents lists available at ScienceDirect
Organic Electronics journal homepage: www.elsevier.com/locate/orgel
Analysis of alternating current driven electroluminescence in organic light emitting diodes: A comparative study Le Zhang a,b, Hajime Nakanotani a, Kou Yoshida a, Chihaya Adachi a,b,⇑ a b
Center for Organic Photonics and Electronics Research (OPERA), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan International Institute for Carbon Neutral Energy Research (WPI-I2CNER), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan
a r t i c l e
i n f o
Article history: Received 13 February 2014 Received in revised form 16 April 2014 Accepted 9 May 2014 Available online 28 May 2014 Keywords: AC-driven double-insulated OLED Space charges Phase shift EL saturation
a b s t r a c t The alternating current (AC) responses of double-injection and double-insulated organic light-emitting diodes (OLEDs) were investigated and compared. To reveal the electroluminescent (EL) processes in these devices, the AC voltage and frequency dependence of the EL intensity and capacitive current were studied in the time domain with a focus on phase difference analysis. It was found that the voltage-dependent transit time and frequencydependent carrier distribution were important for the AC-driven performance of the double-injection OLEDs. In contrast, although the double-insulated OLEDs shared some similarities with the double-injection OLEDs, they had some unique characteristics, which were the absence of resistive current and phase shift of EL profiles. It was revealed that the EL in the double-insulated OLEDs was driven by the displacement current generated by the ionization of the doped layers, which, however, formed space charge regions and undermined the EL emission. The space charge redistributed the electric field across the devices after the initiation of EL, making the EL maintain for a limited time interval. This effect was significant under low frequency and high AC voltage. Comparing the phase difference between both devices, it was indicated that the space charge effect was responsible for the observed EL phase shift and the asymmetric EL profiles at low frequency and high AC voltage in the double-insulated OLEDs. The proposed model was also of help to understand the EL saturation phenomena with AC frequency and voltage in those devices. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Organic light emitting diodes (OLEDs) have advanced rapidly during the last several decades from academic studies to real applications [1–3]. To further explore the ability of OLEDs, new operating mechanisms are expected. Currently, alternating current (AC)- or field-driven organic electro-luminescent (EL) devices are drawing much attention because of their unique characteristics such as less
⇑ Corresponding author at: Center for Organic Photonics and Electronics Research (OPERA), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan. Tel.: +81 92 802 6920; fax: +81 92 802 6921. E-mail address: [email protected] (C. Adachi). http://dx.doi.org/10.1016/j.orgel.2014.05.009 1566-1199/Ó 2014 Elsevier B.V. All rights reserved.
dependence on metal electrodes than traditional OLEDs, easy encapsulation, and possible operation under AC power lines [4–9]. In fact, the idea of AC-driven OLEDs was firstly proposed to make use of the reverse bias region [4,10]. The devices used were common double-injection type OLEDs. Although traditional OLEDs were initially believed to emit light only under forward bias conditions, after introducing two polyaniline layers, light emission was observed under both forward and reverse bias conditions [11–13]. Meanwhile, to avoid the problems associated with low-work-function metal electrodes, organic thin-film electroluminescent devices with a double-insulated structure were fabricated [5,14–16]. Although these devices worked under AC-driven
1816
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
conditions, the driving voltages were rather high. Recently, another structure for AC-driven OLEDs including one injection electrode and one blocking electrode, i.e., a singleinsulated structure, has been reported [7,17,18] However, this structure requires careful considerations of the injection electrode [19]. Because of their unique properties, we focused on OLEDs with double-insulated structures in this study. As to double-insulated OLEDs, there are two important issues. One is the charge carrier generation process, and the other is the space charge effect. Several methods have been proposed for the charge carrier generation process, such as by introducing nanoparticles (NPs), colloidal quantum dots, or doping layers (p–i–n structure) [5,8,9]. In each case, there is a corresponding operating mechanism. For the p–i–n structure, it is suggested that charge carriers (holes and electrons) are generated in the charge generation layers (p-doped and n-doped regions) when the forward AC voltage exceeds the turn-on voltage. Then, the generated carriers are driven by the AC voltage, leading to bipolar carrier transport and radiative recombination. The ionized dopants can be neutralized under the reverse bias condition through a proposed tunneling mechanism [9]. In this way, the electroluminescence (EL) is repeatedly emitted under the forward bias condition. For the p–i–n structure, the carrier concentration can be tuned by the doping concentration [9]. In addition, the operating voltage can be greatly reduced by using insulators with a high dielectric constant [8,20]. Because double-insulated OLEDs are not injection-type, the ionized charge generation layers form space charge regions during the forward bias condition (charge generation process), which redistribute the electric field across the device [8,21]. Besides the ionized charge generation regions, the trapped charges also serve as an important source of the space charge [22–24]. Before the initiation of EL, the ionized dopants are electronically screened by the mobile carriers. The applied AC voltage is mainly to drive the mobile carriers to counter electrodes. However, once the EL occurs, the amount of mobile carriers decreases, leading to an additional electric field generated by the unscreened space charges. The direction of the generated electric field is opposite to the applied AC field, resulting in a decrease of the voltage across the emitting layer and sharp electroluminescent profiles in the time domain [8,9,20,21]. In other words, under the forward bias condition, there are four processes, i.e., ionization of the doped layers, double-carrier transport, recombination and space charge field establishment. The space charge effect will be discussed in the following sections. Although there are increasing studies on the double-insulated OLEDs, the above two issues have still not been completely addressed [8,9,20,25,26]. For a clear understanding of the operating mechanism of double-insulated OLEDs, a comparison of double-injection and double-insulated OLEDs is expected. In the present study, the characteristics of double-injection and double-insulated OLEDs under AC-driven conditions were investigated in a wide frequency region and compared with a focus on their AC frequency and amplitude dependence. Experimental results suggested that although both devices shared some behavior, the
double-insulated structure possesses several unique characteristics that are attributed to the limited charge generation ability and space charge effect. In addition, it was indicated that the phase difference analysis, which was overlooked in the previous studies, was of help for a clear understanding of the operating mechanism. 2. Experiments The devices used here have a homo-junction triplelayer p–i–n diode structure. The double-injection device (Device A) was fabricated as follows. A patterned indium tin oxide (ITO) substrate was used as a transparent anode after cleaning and UV/ozone treatment. Then, a hole injection layer of MoO3 (20 wt.%) doped 4,40 -bis[(N-carbazole) styryl]biphenyl (BSB-Cz) with a thickness of 70 nm was evaporated on the ITO surface [24,27]. Next, a neat BSBCz layer with a thickness of 50 nm was deposited. Subsequently, a 50 nm thick layer of Cs (20 wt.%) doped in BSB-Cz was deposited as an electron injection layer. Finally, an Ag electrode with a thickness of 100 nm was deposited as the cathode. During the evaporation processes, the vacuum pressure was kept to about 104 Pa. The film thickness was monitored using a quartz crystal microbalance and the evaporation rate was about 1 Å/s. For the double-insulated device (Device B), the electrodes and organic layers were exactly the same as those of Device A. The carrier injection layers in Device A served as carrier generation layers in Device B. Hafnium oxide (HfO2) insulating layers were prepared by rf-sputtering from a HfO2 target with a diameter of 5 cm under argon atmosphere with a flow of 50 sccm. The sputtering power was set to about 50 W, resulting in an evaporation rate of about 2.1 nm/min. The thickness of the HfO2 insulating layers was about 100 nm. All devices were encapsulated in dry nitrogen atmosphere shortly after fabrication. The structures of both devices are shown as the insets in Fig. 1. For the AC characterization, a wave function generator (NF, W1974) was used to produce sinusoidal wave with frequency from 1 Hz to 1 MHz, which was amplified by a bipolar high-speed amplifier (NF, HSA4101) from 0 to 75 V. In the study, the device characteristics were measured with respect to the AC voltage amplitude, i.e., 0.5 Vpp, where Vpp is the peak-to-peak voltage. The intensity of electroluminescence (EL, VEL) was measured using a photo-multiplier tube (Hamamatsu, R925). During experiments, the EL intensity was monitored with respect to AC amplitude and frequency. In addition, the current flowing V through the devices (IRs ¼ RRss ) was measured using a 50ohm resistor (Rs) connected in series with the devices. All signals were displayed in an oscilloscope (Agilent, DSO5034A). 3. Results and discussion 3.1. Double-injection OLEDs (Device A) Fig. 1(a) shows the dependence of EL intensity on the applied AC voltage amplitude (Vac = 0.5 Vpp) at an AC frequency of 1 kHz for Device A. For comparison, its current
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
1817
Fig. 2. Frequency-dependent EL intensity of Devices A and B at various AC voltage amplitudes. The frequency at which the EL intensity began to decrease is defined as the characteristic frequency. The capacitive component of the current flowing through the devices is also shown, which is linearly proportional to the AC frequency. It should be noted that the curves of EL intensity have been shifted to allow clear comparison.
Fig. 1. AC voltage amplitude (Vac) dependent EL intensity and current density–voltage–luminance relationship under DC driven condition of (a) Device A (f ¼ 1 kHz) and (b) Device B (f ¼ 10 kHz), respectively. For Device B, the current density and luminance under DC bias were too weak to observe. The insets show device structures.
density–voltage-luminance relationship under the DC bias condition was also illustrated. EL intensity under AC driven condition increased rapidly above a turn-on voltage around 3 V, which was similar to that under DC bias condition [24]. Nevertheless, the luminance under AC driven condition was lower than that under DC bias condition. The luminance was measured using an integrating sphere equipped with a spectrometer (Hamamatsu Photonics, PMA-12). The EL intensity of Device A varied with both AC amplitude and frequency, which was clearly shown in Fig. 2. Since electrons and holes can be smoothly injected from the electrodes into Device A, the conduction was transport limited rather than injection limited. In other words, the conduction mechanisms can be described well by the existing theories for DC conduction [24,27–30]. Under AC-driven conditions, there was competition between the carrier transit time (str ) and AC characteristic time (sac ¼ 1=f ). In the lower frequency region (str < sac ), the injected holes and electrons have enough time to be transported across the active layer and recombine with each other to emit light. Assuming a narrow distribution of carrier mobility, the EL intensity should be independent
on AC frequency in this region [12,13,31]. The decrease of EL intensity at 5 V with frequency was attributed to the voltage dependent carrier mobility distribution [24]. With the increase of AC frequency (str > sac ), the amount of carriers that could travel across the active layer to form excitons began to decrease, leading to a reduction in EL intensity. In other words, injected carrier distribution in space exhibited a frequency dependence [31]. In addition, because the transit time was directly related to the voltage, the characteristic frequency shifted to higher frequency under higher AC voltage, as shown in Fig. 2. The characteristic frequency refers to the frequency over which EL intensity begins to decrease. The mobility of the slower carrier (electrons) can be estimated from Fig. 2 by the relationship:
le ¼
L2 str V ac
ð1Þ
Here, L is the thickness of the active layer, str is the transit time, which is taken as the reciprocal of the characteristic frequency, and V ac is the amplitude of the AC voltage. The estimated electron mobility was about 106 cm2 V1 s1, which was consistent with that estimated previously by impedance spectroscopy [24]. Depending on the frequency, the profile of electroluminescence varied correspondingly, which were shown in the time domain in Fig. 3. At lower frequencies (< 100 Hz), the current has exactly the same phase with EL, implying a resistive characteristic. In other words, the EL was caused by the injected carriers. At the maximum AC voltage, VEL and V Rs reached to their maximum values simultaneously. EL with a peak shape was only observed under the forward bias conditions. With increasing frequency, besides the inphase resistive component, a capacitive component was also observed. The magnitude of this capacitive component increased with frequency, as demonstrated in Fig. 2. In contrast, the magnitude of the resistive component
1818
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 3. Time-resolved EL intensity (VEL) and voltage drop across the resistor (V Rs ) of Device A under Vac = 4 V at frequencies of (a) 1 Hz, (b) 1 kHz, (c) 10 kHz, (d) 100 kHz. It should be noted that at lower frequency (< 100 Hz), V Rs coincided completely with VEL. All data were normalized to their maximum values to allow clear comparison.
decreased gradually with frequency. Therefore, the total current flowing through the device could be expressed as
I ¼ Ic þ Id ¼ GV ac þ C
dV ac dt
ð2Þ
Here, Ic and Id are the conduction and displacement currents, respectively. G is the admittance and C is the capacitance of Device A. From the slope of the displacement current with respect to frequency shown in Fig. 2, the device capacitance was obtained as about 457.3 pF, which was lower than the geometric capacitance of 728 pF. The geometric capacitance was estimated with a relative permittivity of about 3.5, electrode area of 4 mm2 and thickness of 170 nm. In the lower frequency region, the EL profiles were located exactly at the peak positions of AC voltages. However, when the frequency was increased (f > 20 kHz), the EL profiles lagged behind Vac peaks, indicating a response delay caused by the finite carrier mobility [12,13,32]. For a deeper understanding of the behavior of Device A, it is helpful to look at the phase differences, which are shown in Fig. 4. The definition of the phase difference is illustrated as the inset of Fig. 4. In our definitions, all the phases are defined with respect to the onset of AC voltage. The phase difference [D/ðV EL Þ] between EL and AC voltage was kept about 90° below the characteristic frequency, over which the phase began to increase, i.e., phase lag. In other words, under the characteristic frequency, the maximum EL intensity coincided with the maximum resistive current and Vac. However, the maximum EL intensity lagged behind the maximum Vac above the characteristic frequency. The characteristic frequency varied with AC voltage, which was consistent with the previous analysis. In contrast, the phase difference [D/ðV Rs Þ] between the capacitive current and AC voltage was slightly lower than 90°, implying the coexistence of a resistive component.
Fig. 4. The frequency-dependent phase differences of Devices A and B under various AC voltages. The definitions of the phase differences [D/ðV EL Þ and D/ðV Rs Þ] are shown as the inset. The horizontal dashed line indicates the 90° position. Dt is the duration of EL and T is the period of AC voltage. The absence of the phase difference data in the lower frequency region (< 100 Hz) is caused by the weak capacitive component of the V Rs signal.
The mixed properties of the current may be responsible for the underestimated capacitance. Above the characteristic frequency, the phase began to decrease, implying an increasing resistive property. Finally, it is of interest to investigate the AC voltagedependent EL emission profiles of Device A in the time domain, which are presented in Fig. 5(a)–(c). The peak position was not affected by AC amplitude for Device A, i.e., the maximum EL intensity coincided with the maximum AC voltage. In addition, the EL profiles were symmetric. With increasing AC amplitude, the width of the EL
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 5. Time-resolved AC voltage dependent EL profiles at 1 Hz for Device A at (a) 3 V, (b) 4 V and (c) 5 V and Device B at (d) 10 V, (e) 15 V and (f) 20 V. The vertical arrows indicate the turn-on of EL emission and the horizontal arrows indicate the AC voltage at maximum EL intensity (V max ).
emission (Dt) increased, which was consistent with the conduction mechanism discussed above [10]. 3.2. Double-insulated OLEDs (Device B) For the double-insulated device, the carriers are generated from the doped layers by the ionization process under the forward bias condition. The generated carriers transport along the direction of AC voltage and then recombine to emit light. Differing from the double-injection OLEDs, depleted space charge regions are formed during the charge generation process for the double-insulated OLEDs, which can be neutralized under the reverse bias condition through a proposed tunneling process [9]. It should be noted that the space charge region plays a deterministic role in the analysis of the EL characteristics of the double-insulated OLEDs, which will be discussed below. The AC-voltage dependent EL intensity of Device B is illustrated in Fig. 1(b) at a frequency of 10 kHz. The EL intensity of Device B was much lower than that of Device A, and the turn-on voltage at 0.1 cd/m2 of Device B was much higher. Although its luminance was a little lower than the reported values [20,25], it should be noted that (1) the thickness of active and insulating layers were larger in our device; (2) visible light (k ¼ 480 nm) was uniformly emitted from the device surface under AC-driven conditions but no detectable emission was observed under DC-driven conditions. The voltage across the organic layers (Vorg) was estimated using the equation:
1þ
org HfO
2
dHfO2 V org ¼ V ac dorg
ð3Þ
Here, org ðdorg Þ and HfO2 ðdHfO2 Þ are the dielectric constants (layer thickness) of the organic and insulating layers, respectively. As a rough estimation with org ð3:5Þ and HfO2 ð21Þ, Vorg is about 0.84 Vac. To further improve the device performance, materials with high dielectric constant and excellent insulating properties are required. Here, it should be noted that for simplicity, the effects of
1819
depleted space charge regions are not involved in the above equation. The frequency-dependent EL intensity of Device B is presented in Fig. 2. The frequency dependence can be divided into three regions. At low frequency (< 100 Hz), the EL intensity increased with frequency, implying that it was driven by displacement current rather than conduction current. Within the intermediate-frequency region (102 < f < 104 Hz), the EL intensity became saturated, and then decreased slightly when the frequency exceeded 10 kHz. Similar to Device A, the behavior in higher frequency region can be understood by considering the relationship between transit time and AC frequency dependent carrier distribution. Nevertheless, the reasons for the increase of EL intensity in the low-frequency region need further examination. The capacitive currents exhibited a linear dependence on AC frequency, similar to Device A. The current of Device B was one order of magnitude higher than that of Device A because of its higher capacitance and Vac. Using the method described previously, the capacitance of Device B was estimated to be 1.36 nF. The larger capacitance was attributed to the decrease of effective organic layer thickness after rf sputtering [9,33,34]. These results show that although the frequency dependence of the EL intensity and current of the two devices shared some behavior, they were different in the low-frequency region. Fig. 6 illustrates the time-resolved AC responses of Device B at various frequencies. Similar to Device A, the profiles of EL intensity exhibited a peak like shape. However, there were several unique characteristics. Firstly, the current was totally capacitive with a phase difference of 90°, implying that the EL was driven completely by the displacement current. Secondly, the phase difference between VEL and Vac was smaller than 90° in the low frequency region (< 100 Hz). In addition, it showed an asymmetric profile at lower frequency. A small phase difference has been observed in various double-insulated devices, but it is not well understood [17–21,25]. However, in the higher frequency region, the phase difference [D/ðV EL Þ] returned to 90° and even over 90° at higher frequency (> 104 Hz). The phase differences of Devices A and B were shown in Fig. 4. For both devices, the maximum EL intensity occurred when the phase difference between VEL and Vac was 90°. In addition, except in the low-frequency region, the phase differences [D/ðV EL Þ and D/ðV Rs Þ] in Devices A and B exhibited similar behavior. To better understand the difference between the two devices at low frequency, it is necessary to obtain their EL profiles, which are illustrated in Fig. 5. EL intensities varied in phase with AC voltages for Device A (Fig. 5(a)– (c)). Increased AC voltage did not affect the positions of the maximum EL intensity, it only extended the width of the EL profiles. Furthermore, the profile of EL intensity was symmetric with respect to its maximum. In contrast, for Device B (Fig. 5(d)–(f)), with increasing of AC voltage, the EL profiles shifted gradually towards the left and then remained still. In addition, the profiles of EL intensity changed from symmetric to asymmetric, and was mainly distributed in the first quarter-period of AC voltage. To understand this phenomenon, the space charge effect
1820
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 6. Time-resolved EL and V Rs characteristics of Device B under V ac ¼ 17:5 V at frequencies of (a) 1 Hz, (b) 100 Hz, (c) 10 kHz, (d) 100 kHz. It should be noted that in the lower frequency region (< 100 Hz), V Rs was too small to observe. All data were normalized to their maximum values to allow a clear comparison.
should be considered. During the first quarter-period of AC voltage, the AC voltage gradually increased, resulting in mobile carriers (electrons and holes) and immobile ionized space charge layers. The mobile carriers travelled along the direction of the applied AC voltage, leading to exciton formation and radiative recombination. Before the initiation of EL, the depleted space charges were electronically screened by the generated mobile carriers. The applied AC voltage was to drive the carriers towards the counter electrodes. Nevertheless, once the EL was occurred, the amount of mobile carriers decreased. Consequently, an additional electric field was established by the unscreened space charges in both doped layers. The direction of the established field was opposite to the applied AC field, leading to a decrease of the net electric field across the active layer. The process was schematically illustrated as the inset in Fig. 7. For the double-insulated device, the space charge effect made the EL emission only remained for a limited time interval after it started. Similar to Device A, with the increase of AC voltage in Device B, the width of EL intensity should become wider. However, because of the space charge effect in Device B, once EL started, it only lasted for a limited time period. When the AC voltage was smaller than the critical voltage, the EL peak was located at the peak position of AC voltage, as shown in Fig. 5(d). The voltage at which EL reached to its maximum (Vmax) was same with the amplitude of AC voltage (Vac). However, once AC voltage was increased above the critical voltage, the EL intensity was pinned at the critical voltage and lasted for a limited time interval, resulting in the observed phase shift phenomenon. The AC voltage at which the EL began to saturate was defined as the critical voltage (Vc) as shown in Fig. 7. At one AC frequency, the space charge effect was only significant when Vac > Vc. The asymmetric EL profiles also implied the influence from space charge effect, which caused a different EL relaxation process. For the establishment process of EL when V < Vc,
Fig. 7. AC-voltage dependent Vmax (voltage at maximum EL intensity) and EL phase difference [D/ðV EL Þ] at various frequencies. The critical voltage (Vc) was defined as the voltage, after which Vmax was saturated. The insets showed the proposed model for the space charge effect. Qm, Qd, and Qf were the charges on metal electrodes, ionized dopants and free mobile carriers, respectively. + and indicated the polarity of the charges. Eac and Ein are the applied AC electric field and generated space charge field, respectively.
EL intensity was only determined by the carriers generated by the ionization of doped layers. In contrast, after the initiation of EL, it was a space charge modulated process [23]. The critical voltage depended on the AC voltage and frequency, as depicted in Fig. 7. In Fig. 7, Vac dependence of Vmax and D/ðV EL Þ was shown under various AC frequency. It can be seen that at lower frequency (1 Hz), the saturation behavior was well observed. Below Vc, Vmax was equal to AC voltage amplitude, i.e., EL intensity was maximum at AC amplitude. In this case, the space charge effect was insignificant. However, when Vac increased above Vc, Vmax was pinned at Vc. Although this behavior became difficult
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
to distinguish under higher AC frequency (10 and 100 Hz), the increase of Vmax with AC amplitude gradually slowed down, implying that the critical voltage shifted to higher voltage outside of the measurement range. A higher frequency corresponded to a higher critical voltage and vice versa. In addition, at one AC voltage, EL phase approached to 90° with increasing frequency, which was consistent with Fig. 4. To further characterize the space charge effect, simulation is expected, which will be our next work. 4. Conclusions In conclusion, the AC responses of double-injection and double-insulated OLEDs were investigated and compared. For the double-injection OLEDs, the voltage-dependent transit time and frequency dependent carrier distribution were responsible for the observed phenomena. Although capacitive current was observed in the double-injection device, especially in the high-frequency region, the main contribution to the EL was attributed to the in-phase resistive current. On the other hand, for the double-insulated OLEDs, although they shared some characteristics with the double-injection ones, there were two differences between the two devices. One was the absence of resistive component and the other was the small EL phase difference (< 90 ) in the low-frequency region at high AC voltage. These phenomena were explained by considering the space charge effect, which caused the EL to last for a limited time interval. The phase difference analysis corresponded well with other results, but from a different standpoints with new information. Maximum EL intensity was found only when the EL phase difference was 90° in both devices, i.e., maximum EL intensity coincided with maximum AC voltage. The destroyed EL profiles due to the space charge effects were suggested to be responsible for the much lower EL intensity than that of the doubleinjection OLEDs and the EL intensity saturation at higher AC amplitude. To further improve device performance under AC driving conditions, on one hand, the charge generation ability should be enhanced with balanced carriers generation. On the other hand, the space charge effect should be carefully considered by optimizing the device structure and operating conditions (AC voltage and frequency). Acknowledgments This work was supported by a Grant-in-aid from the Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST) and the International Insti-
1821
tute for Carbon Neutral Energy Research (WPI-I2CNER) sponsored by the Ministry of Education, Culture, Sports, Science and Technology of Japan. References [1] S.R. Forrest, Nature 428 (2004) 911. [2] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 492 (2012) 234. [3] W. Brütting, C. Adachi, Physics of Organic Semiconductors 2nd, Wiley-VCH, Berlin, 2012. [4] Y.Z. Wang, D.D. Gebler, L.B. Lin, J.W. Blatchford, S.W. Jessen, H.L. Wang, A.J. Epstein, Appl. Phys. Lett. 68 (1996) 894. [5] T. Tsutsui, S. Lee, K. Fujita, Appl. Phys. Lett. 85 (2004) 2382. [6] T. Yamao, Y. Shimizu, K. Terasaki, S. Hotta, Adv. Mater. 20 (2008) 4109. [7] J. Sung, Y.S. Choi, S.J. Kang, S.H. Cho, T. Lee, C. Park, Nano. Lett. 11 (2011) 966. [8] V. Wood, M.J. Panzer, D. Bozyigit, Y. Shirasaki, I. Rousseau, S. Geyer, M.G. Bawendi, V. Bulovic´, Nano. Lett. 11 (2011) 2927. [9] A. Perumal, M. Fröbel, S. Gorantla, T. Gemming, B. Lüssem, J. Eckert, K. Leo, Adv. Funct. Mater. 22 (2012) 210. [10] Z. Yang, B. Hu, F.E. Karasz, Macromolecules 28 (1995) 6151. [11] H.L. Wang, A.G. MacDiarmid, Y.Z. Wang, D.D. Gebler, A.J. Epstein, Synth. Met. 78 (1996) 33. [12] A.J. Pal, R. Österbacka, K.-M. Källman, H. Stubb, Appl. Phys. Lett. 70 (1997) 2022. [13] R. Österbacka, A.J. Pal, K.-M. Källman, H. Stubb, J. Appl. Phys. 83 (1998) 1748. [14] X.L. Xu, X.H. Chen, Y.B. Hou, Z. Xu, X.H. Yang, S.G. Yin, Z.J. Wang, X.R. Xu, S.P. Lau, B.K. Tay, Chem. Phys. Lett. 325 (2000) 420. [15] S.Y. Yang, L. Qian, F. Teng, Z. Xu, X.R. Xu, J. Appl. Phys. 97 (2005) 126101. [16] S. Lee, K. Fujita, T. Tsutsui, Jpn. J. Appl. Phys. 44 (2005) 6607. [17] J. Brückner, N. Christ, O. Bauder, C. Gärtner, M. Seyfried, F. Glückler, U. Lemmer, M. Gerken, Appl. Phys. Lett. 96 (2010) 041107. [18] Y. Chen, G.M. Smith, E. Loughman, Y. Li, W. Nie, D.L. Carroll, Org. Electron. 14 (2013) 8. [19] Y. Xia, Y. Chen, G.M. Smith, Y. Li, W. Huang, D.L. Carroll, Appl. Phys. Lett. 102 (2013) 253302. [20] A. Perumal, B. Lüssem, K. Leo, Appl. Phys. Lett. 100 (2012) 103307. [21] D. Bozyigit, V. Wood, Y. Shirasaki, V. Bulovic, J. Appl. Phys. 102 (2012) 113701. [22] M.A. Lampert, P. Mark, Current Injection in Solids, Academic Press, New York, 1970. [23] L. Zhang, D. Taguchi, T. Manaka, M. Iwamoto, J. Appl. Phys. 110 (2011) 033715. [24] L. Zhang, H. Nakanotani, C. Adachi, Appl. Phys. Lett. 103 (2013) 093301. [25] M. Fröbel, A. Perumal, T. Schwab, M.C. Gather, B. Lüssem, K. Leo, Org. Electron. 14 (2013) 809. [26] A. Perumal, B. Lüssem, K. Leo, Org. Electron. 13 (2012) 1589. [27] Y. Setoguchi, C. Adachi, J. Appl. Phys. 108 (2010) 064516. [28] D. Taguchi, S. Inoue, L. Zhang, J. Li, M. Weis, T. Manaka, M. Iwamoto, J. Phys. Chem. Lett. 1 (2010) 803. [29] D. Taguchi, L. Zhang, J. Li, M. Weis, T. Manaka, M. Iwamoto, J. Phys. Chem. C 114 (2010) 15136. [30] W. Brütting, S. Berleb, A.G. Muckl, Org. Electron. 2 (2001) 1. [31] S. Paul, P. Pal, A.J. Pal, G.B. Talapatra, J. Appl. Phys. 94 (2003) 4357. [32] S. Das, A. Chowdhury, A.J. Pal, Phys. Stat. Sol. (a) 185 (2001) 383. [33] D. Vaufrey, M.B. Khalifa, J. Tardy, C. Ghica, M.G. Blanchin, C. Sandu, J.A. Roger, Semicond. Sci. Technol. 18 (2003) 253. [34] H. Fujimoto, T. Miyayama, N. Sanada, C. Adachi, Org. Electron. 14 (2013) 2994.
Contents lists available at ScienceDirect
Organic Electronics journal homepage: www.elsevier.com/locate/orgel
Analysis of alternating current driven electroluminescence in organic light emitting diodes: A comparative study Le Zhang a,b, Hajime Nakanotani a, Kou Yoshida a, Chihaya Adachi a,b,⇑ a b
Center for Organic Photonics and Electronics Research (OPERA), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan International Institute for Carbon Neutral Energy Research (WPI-I2CNER), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan
a r t i c l e
i n f o
Article history: Received 13 February 2014 Received in revised form 16 April 2014 Accepted 9 May 2014 Available online 28 May 2014 Keywords: AC-driven double-insulated OLED Space charges Phase shift EL saturation
a b s t r a c t The alternating current (AC) responses of double-injection and double-insulated organic light-emitting diodes (OLEDs) were investigated and compared. To reveal the electroluminescent (EL) processes in these devices, the AC voltage and frequency dependence of the EL intensity and capacitive current were studied in the time domain with a focus on phase difference analysis. It was found that the voltage-dependent transit time and frequencydependent carrier distribution were important for the AC-driven performance of the double-injection OLEDs. In contrast, although the double-insulated OLEDs shared some similarities with the double-injection OLEDs, they had some unique characteristics, which were the absence of resistive current and phase shift of EL profiles. It was revealed that the EL in the double-insulated OLEDs was driven by the displacement current generated by the ionization of the doped layers, which, however, formed space charge regions and undermined the EL emission. The space charge redistributed the electric field across the devices after the initiation of EL, making the EL maintain for a limited time interval. This effect was significant under low frequency and high AC voltage. Comparing the phase difference between both devices, it was indicated that the space charge effect was responsible for the observed EL phase shift and the asymmetric EL profiles at low frequency and high AC voltage in the double-insulated OLEDs. The proposed model was also of help to understand the EL saturation phenomena with AC frequency and voltage in those devices. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Organic light emitting diodes (OLEDs) have advanced rapidly during the last several decades from academic studies to real applications [1–3]. To further explore the ability of OLEDs, new operating mechanisms are expected. Currently, alternating current (AC)- or field-driven organic electro-luminescent (EL) devices are drawing much attention because of their unique characteristics such as less
⇑ Corresponding author at: Center for Organic Photonics and Electronics Research (OPERA), Kyushu University, 744 Motooka, Nishi, Fukuoka 819-0395, Japan. Tel.: +81 92 802 6920; fax: +81 92 802 6921. E-mail address: [email protected] (C. Adachi). http://dx.doi.org/10.1016/j.orgel.2014.05.009 1566-1199/Ó 2014 Elsevier B.V. All rights reserved.
dependence on metal electrodes than traditional OLEDs, easy encapsulation, and possible operation under AC power lines [4–9]. In fact, the idea of AC-driven OLEDs was firstly proposed to make use of the reverse bias region [4,10]. The devices used were common double-injection type OLEDs. Although traditional OLEDs were initially believed to emit light only under forward bias conditions, after introducing two polyaniline layers, light emission was observed under both forward and reverse bias conditions [11–13]. Meanwhile, to avoid the problems associated with low-work-function metal electrodes, organic thin-film electroluminescent devices with a double-insulated structure were fabricated [5,14–16]. Although these devices worked under AC-driven
1816
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
conditions, the driving voltages were rather high. Recently, another structure for AC-driven OLEDs including one injection electrode and one blocking electrode, i.e., a singleinsulated structure, has been reported [7,17,18] However, this structure requires careful considerations of the injection electrode [19]. Because of their unique properties, we focused on OLEDs with double-insulated structures in this study. As to double-insulated OLEDs, there are two important issues. One is the charge carrier generation process, and the other is the space charge effect. Several methods have been proposed for the charge carrier generation process, such as by introducing nanoparticles (NPs), colloidal quantum dots, or doping layers (p–i–n structure) [5,8,9]. In each case, there is a corresponding operating mechanism. For the p–i–n structure, it is suggested that charge carriers (holes and electrons) are generated in the charge generation layers (p-doped and n-doped regions) when the forward AC voltage exceeds the turn-on voltage. Then, the generated carriers are driven by the AC voltage, leading to bipolar carrier transport and radiative recombination. The ionized dopants can be neutralized under the reverse bias condition through a proposed tunneling mechanism [9]. In this way, the electroluminescence (EL) is repeatedly emitted under the forward bias condition. For the p–i–n structure, the carrier concentration can be tuned by the doping concentration [9]. In addition, the operating voltage can be greatly reduced by using insulators with a high dielectric constant [8,20]. Because double-insulated OLEDs are not injection-type, the ionized charge generation layers form space charge regions during the forward bias condition (charge generation process), which redistribute the electric field across the device [8,21]. Besides the ionized charge generation regions, the trapped charges also serve as an important source of the space charge [22–24]. Before the initiation of EL, the ionized dopants are electronically screened by the mobile carriers. The applied AC voltage is mainly to drive the mobile carriers to counter electrodes. However, once the EL occurs, the amount of mobile carriers decreases, leading to an additional electric field generated by the unscreened space charges. The direction of the generated electric field is opposite to the applied AC field, resulting in a decrease of the voltage across the emitting layer and sharp electroluminescent profiles in the time domain [8,9,20,21]. In other words, under the forward bias condition, there are four processes, i.e., ionization of the doped layers, double-carrier transport, recombination and space charge field establishment. The space charge effect will be discussed in the following sections. Although there are increasing studies on the double-insulated OLEDs, the above two issues have still not been completely addressed [8,9,20,25,26]. For a clear understanding of the operating mechanism of double-insulated OLEDs, a comparison of double-injection and double-insulated OLEDs is expected. In the present study, the characteristics of double-injection and double-insulated OLEDs under AC-driven conditions were investigated in a wide frequency region and compared with a focus on their AC frequency and amplitude dependence. Experimental results suggested that although both devices shared some behavior, the
double-insulated structure possesses several unique characteristics that are attributed to the limited charge generation ability and space charge effect. In addition, it was indicated that the phase difference analysis, which was overlooked in the previous studies, was of help for a clear understanding of the operating mechanism. 2. Experiments The devices used here have a homo-junction triplelayer p–i–n diode structure. The double-injection device (Device A) was fabricated as follows. A patterned indium tin oxide (ITO) substrate was used as a transparent anode after cleaning and UV/ozone treatment. Then, a hole injection layer of MoO3 (20 wt.%) doped 4,40 -bis[(N-carbazole) styryl]biphenyl (BSB-Cz) with a thickness of 70 nm was evaporated on the ITO surface [24,27]. Next, a neat BSBCz layer with a thickness of 50 nm was deposited. Subsequently, a 50 nm thick layer of Cs (20 wt.%) doped in BSB-Cz was deposited as an electron injection layer. Finally, an Ag electrode with a thickness of 100 nm was deposited as the cathode. During the evaporation processes, the vacuum pressure was kept to about 104 Pa. The film thickness was monitored using a quartz crystal microbalance and the evaporation rate was about 1 Å/s. For the double-insulated device (Device B), the electrodes and organic layers were exactly the same as those of Device A. The carrier injection layers in Device A served as carrier generation layers in Device B. Hafnium oxide (HfO2) insulating layers were prepared by rf-sputtering from a HfO2 target with a diameter of 5 cm under argon atmosphere with a flow of 50 sccm. The sputtering power was set to about 50 W, resulting in an evaporation rate of about 2.1 nm/min. The thickness of the HfO2 insulating layers was about 100 nm. All devices were encapsulated in dry nitrogen atmosphere shortly after fabrication. The structures of both devices are shown as the insets in Fig. 1. For the AC characterization, a wave function generator (NF, W1974) was used to produce sinusoidal wave with frequency from 1 Hz to 1 MHz, which was amplified by a bipolar high-speed amplifier (NF, HSA4101) from 0 to 75 V. In the study, the device characteristics were measured with respect to the AC voltage amplitude, i.e., 0.5 Vpp, where Vpp is the peak-to-peak voltage. The intensity of electroluminescence (EL, VEL) was measured using a photo-multiplier tube (Hamamatsu, R925). During experiments, the EL intensity was monitored with respect to AC amplitude and frequency. In addition, the current flowing V through the devices (IRs ¼ RRss ) was measured using a 50ohm resistor (Rs) connected in series with the devices. All signals were displayed in an oscilloscope (Agilent, DSO5034A). 3. Results and discussion 3.1. Double-injection OLEDs (Device A) Fig. 1(a) shows the dependence of EL intensity on the applied AC voltage amplitude (Vac = 0.5 Vpp) at an AC frequency of 1 kHz for Device A. For comparison, its current
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
1817
Fig. 2. Frequency-dependent EL intensity of Devices A and B at various AC voltage amplitudes. The frequency at which the EL intensity began to decrease is defined as the characteristic frequency. The capacitive component of the current flowing through the devices is also shown, which is linearly proportional to the AC frequency. It should be noted that the curves of EL intensity have been shifted to allow clear comparison.
Fig. 1. AC voltage amplitude (Vac) dependent EL intensity and current density–voltage–luminance relationship under DC driven condition of (a) Device A (f ¼ 1 kHz) and (b) Device B (f ¼ 10 kHz), respectively. For Device B, the current density and luminance under DC bias were too weak to observe. The insets show device structures.
density–voltage-luminance relationship under the DC bias condition was also illustrated. EL intensity under AC driven condition increased rapidly above a turn-on voltage around 3 V, which was similar to that under DC bias condition [24]. Nevertheless, the luminance under AC driven condition was lower than that under DC bias condition. The luminance was measured using an integrating sphere equipped with a spectrometer (Hamamatsu Photonics, PMA-12). The EL intensity of Device A varied with both AC amplitude and frequency, which was clearly shown in Fig. 2. Since electrons and holes can be smoothly injected from the electrodes into Device A, the conduction was transport limited rather than injection limited. In other words, the conduction mechanisms can be described well by the existing theories for DC conduction [24,27–30]. Under AC-driven conditions, there was competition between the carrier transit time (str ) and AC characteristic time (sac ¼ 1=f ). In the lower frequency region (str < sac ), the injected holes and electrons have enough time to be transported across the active layer and recombine with each other to emit light. Assuming a narrow distribution of carrier mobility, the EL intensity should be independent
on AC frequency in this region [12,13,31]. The decrease of EL intensity at 5 V with frequency was attributed to the voltage dependent carrier mobility distribution [24]. With the increase of AC frequency (str > sac ), the amount of carriers that could travel across the active layer to form excitons began to decrease, leading to a reduction in EL intensity. In other words, injected carrier distribution in space exhibited a frequency dependence [31]. In addition, because the transit time was directly related to the voltage, the characteristic frequency shifted to higher frequency under higher AC voltage, as shown in Fig. 2. The characteristic frequency refers to the frequency over which EL intensity begins to decrease. The mobility of the slower carrier (electrons) can be estimated from Fig. 2 by the relationship:
le ¼
L2 str V ac
ð1Þ
Here, L is the thickness of the active layer, str is the transit time, which is taken as the reciprocal of the characteristic frequency, and V ac is the amplitude of the AC voltage. The estimated electron mobility was about 106 cm2 V1 s1, which was consistent with that estimated previously by impedance spectroscopy [24]. Depending on the frequency, the profile of electroluminescence varied correspondingly, which were shown in the time domain in Fig. 3. At lower frequencies (< 100 Hz), the current has exactly the same phase with EL, implying a resistive characteristic. In other words, the EL was caused by the injected carriers. At the maximum AC voltage, VEL and V Rs reached to their maximum values simultaneously. EL with a peak shape was only observed under the forward bias conditions. With increasing frequency, besides the inphase resistive component, a capacitive component was also observed. The magnitude of this capacitive component increased with frequency, as demonstrated in Fig. 2. In contrast, the magnitude of the resistive component
1818
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 3. Time-resolved EL intensity (VEL) and voltage drop across the resistor (V Rs ) of Device A under Vac = 4 V at frequencies of (a) 1 Hz, (b) 1 kHz, (c) 10 kHz, (d) 100 kHz. It should be noted that at lower frequency (< 100 Hz), V Rs coincided completely with VEL. All data were normalized to their maximum values to allow clear comparison.
decreased gradually with frequency. Therefore, the total current flowing through the device could be expressed as
I ¼ Ic þ Id ¼ GV ac þ C
dV ac dt
ð2Þ
Here, Ic and Id are the conduction and displacement currents, respectively. G is the admittance and C is the capacitance of Device A. From the slope of the displacement current with respect to frequency shown in Fig. 2, the device capacitance was obtained as about 457.3 pF, which was lower than the geometric capacitance of 728 pF. The geometric capacitance was estimated with a relative permittivity of about 3.5, electrode area of 4 mm2 and thickness of 170 nm. In the lower frequency region, the EL profiles were located exactly at the peak positions of AC voltages. However, when the frequency was increased (f > 20 kHz), the EL profiles lagged behind Vac peaks, indicating a response delay caused by the finite carrier mobility [12,13,32]. For a deeper understanding of the behavior of Device A, it is helpful to look at the phase differences, which are shown in Fig. 4. The definition of the phase difference is illustrated as the inset of Fig. 4. In our definitions, all the phases are defined with respect to the onset of AC voltage. The phase difference [D/ðV EL Þ] between EL and AC voltage was kept about 90° below the characteristic frequency, over which the phase began to increase, i.e., phase lag. In other words, under the characteristic frequency, the maximum EL intensity coincided with the maximum resistive current and Vac. However, the maximum EL intensity lagged behind the maximum Vac above the characteristic frequency. The characteristic frequency varied with AC voltage, which was consistent with the previous analysis. In contrast, the phase difference [D/ðV Rs Þ] between the capacitive current and AC voltage was slightly lower than 90°, implying the coexistence of a resistive component.
Fig. 4. The frequency-dependent phase differences of Devices A and B under various AC voltages. The definitions of the phase differences [D/ðV EL Þ and D/ðV Rs Þ] are shown as the inset. The horizontal dashed line indicates the 90° position. Dt is the duration of EL and T is the period of AC voltage. The absence of the phase difference data in the lower frequency region (< 100 Hz) is caused by the weak capacitive component of the V Rs signal.
The mixed properties of the current may be responsible for the underestimated capacitance. Above the characteristic frequency, the phase began to decrease, implying an increasing resistive property. Finally, it is of interest to investigate the AC voltagedependent EL emission profiles of Device A in the time domain, which are presented in Fig. 5(a)–(c). The peak position was not affected by AC amplitude for Device A, i.e., the maximum EL intensity coincided with the maximum AC voltage. In addition, the EL profiles were symmetric. With increasing AC amplitude, the width of the EL
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 5. Time-resolved AC voltage dependent EL profiles at 1 Hz for Device A at (a) 3 V, (b) 4 V and (c) 5 V and Device B at (d) 10 V, (e) 15 V and (f) 20 V. The vertical arrows indicate the turn-on of EL emission and the horizontal arrows indicate the AC voltage at maximum EL intensity (V max ).
emission (Dt) increased, which was consistent with the conduction mechanism discussed above [10]. 3.2. Double-insulated OLEDs (Device B) For the double-insulated device, the carriers are generated from the doped layers by the ionization process under the forward bias condition. The generated carriers transport along the direction of AC voltage and then recombine to emit light. Differing from the double-injection OLEDs, depleted space charge regions are formed during the charge generation process for the double-insulated OLEDs, which can be neutralized under the reverse bias condition through a proposed tunneling process [9]. It should be noted that the space charge region plays a deterministic role in the analysis of the EL characteristics of the double-insulated OLEDs, which will be discussed below. The AC-voltage dependent EL intensity of Device B is illustrated in Fig. 1(b) at a frequency of 10 kHz. The EL intensity of Device B was much lower than that of Device A, and the turn-on voltage at 0.1 cd/m2 of Device B was much higher. Although its luminance was a little lower than the reported values [20,25], it should be noted that (1) the thickness of active and insulating layers were larger in our device; (2) visible light (k ¼ 480 nm) was uniformly emitted from the device surface under AC-driven conditions but no detectable emission was observed under DC-driven conditions. The voltage across the organic layers (Vorg) was estimated using the equation:
1þ
org HfO
2
dHfO2 V org ¼ V ac dorg
ð3Þ
Here, org ðdorg Þ and HfO2 ðdHfO2 Þ are the dielectric constants (layer thickness) of the organic and insulating layers, respectively. As a rough estimation with org ð3:5Þ and HfO2 ð21Þ, Vorg is about 0.84 Vac. To further improve the device performance, materials with high dielectric constant and excellent insulating properties are required. Here, it should be noted that for simplicity, the effects of
1819
depleted space charge regions are not involved in the above equation. The frequency-dependent EL intensity of Device B is presented in Fig. 2. The frequency dependence can be divided into three regions. At low frequency (< 100 Hz), the EL intensity increased with frequency, implying that it was driven by displacement current rather than conduction current. Within the intermediate-frequency region (102 < f < 104 Hz), the EL intensity became saturated, and then decreased slightly when the frequency exceeded 10 kHz. Similar to Device A, the behavior in higher frequency region can be understood by considering the relationship between transit time and AC frequency dependent carrier distribution. Nevertheless, the reasons for the increase of EL intensity in the low-frequency region need further examination. The capacitive currents exhibited a linear dependence on AC frequency, similar to Device A. The current of Device B was one order of magnitude higher than that of Device A because of its higher capacitance and Vac. Using the method described previously, the capacitance of Device B was estimated to be 1.36 nF. The larger capacitance was attributed to the decrease of effective organic layer thickness after rf sputtering [9,33,34]. These results show that although the frequency dependence of the EL intensity and current of the two devices shared some behavior, they were different in the low-frequency region. Fig. 6 illustrates the time-resolved AC responses of Device B at various frequencies. Similar to Device A, the profiles of EL intensity exhibited a peak like shape. However, there were several unique characteristics. Firstly, the current was totally capacitive with a phase difference of 90°, implying that the EL was driven completely by the displacement current. Secondly, the phase difference between VEL and Vac was smaller than 90° in the low frequency region (< 100 Hz). In addition, it showed an asymmetric profile at lower frequency. A small phase difference has been observed in various double-insulated devices, but it is not well understood [17–21,25]. However, in the higher frequency region, the phase difference [D/ðV EL Þ] returned to 90° and even over 90° at higher frequency (> 104 Hz). The phase differences of Devices A and B were shown in Fig. 4. For both devices, the maximum EL intensity occurred when the phase difference between VEL and Vac was 90°. In addition, except in the low-frequency region, the phase differences [D/ðV EL Þ and D/ðV Rs Þ] in Devices A and B exhibited similar behavior. To better understand the difference between the two devices at low frequency, it is necessary to obtain their EL profiles, which are illustrated in Fig. 5. EL intensities varied in phase with AC voltages for Device A (Fig. 5(a)– (c)). Increased AC voltage did not affect the positions of the maximum EL intensity, it only extended the width of the EL profiles. Furthermore, the profile of EL intensity was symmetric with respect to its maximum. In contrast, for Device B (Fig. 5(d)–(f)), with increasing of AC voltage, the EL profiles shifted gradually towards the left and then remained still. In addition, the profiles of EL intensity changed from symmetric to asymmetric, and was mainly distributed in the first quarter-period of AC voltage. To understand this phenomenon, the space charge effect
1820
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
Fig. 6. Time-resolved EL and V Rs characteristics of Device B under V ac ¼ 17:5 V at frequencies of (a) 1 Hz, (b) 100 Hz, (c) 10 kHz, (d) 100 kHz. It should be noted that in the lower frequency region (< 100 Hz), V Rs was too small to observe. All data were normalized to their maximum values to allow a clear comparison.
should be considered. During the first quarter-period of AC voltage, the AC voltage gradually increased, resulting in mobile carriers (electrons and holes) and immobile ionized space charge layers. The mobile carriers travelled along the direction of the applied AC voltage, leading to exciton formation and radiative recombination. Before the initiation of EL, the depleted space charges were electronically screened by the generated mobile carriers. The applied AC voltage was to drive the carriers towards the counter electrodes. Nevertheless, once the EL was occurred, the amount of mobile carriers decreased. Consequently, an additional electric field was established by the unscreened space charges in both doped layers. The direction of the established field was opposite to the applied AC field, leading to a decrease of the net electric field across the active layer. The process was schematically illustrated as the inset in Fig. 7. For the double-insulated device, the space charge effect made the EL emission only remained for a limited time interval after it started. Similar to Device A, with the increase of AC voltage in Device B, the width of EL intensity should become wider. However, because of the space charge effect in Device B, once EL started, it only lasted for a limited time period. When the AC voltage was smaller than the critical voltage, the EL peak was located at the peak position of AC voltage, as shown in Fig. 5(d). The voltage at which EL reached to its maximum (Vmax) was same with the amplitude of AC voltage (Vac). However, once AC voltage was increased above the critical voltage, the EL intensity was pinned at the critical voltage and lasted for a limited time interval, resulting in the observed phase shift phenomenon. The AC voltage at which the EL began to saturate was defined as the critical voltage (Vc) as shown in Fig. 7. At one AC frequency, the space charge effect was only significant when Vac > Vc. The asymmetric EL profiles also implied the influence from space charge effect, which caused a different EL relaxation process. For the establishment process of EL when V < Vc,
Fig. 7. AC-voltage dependent Vmax (voltage at maximum EL intensity) and EL phase difference [D/ðV EL Þ] at various frequencies. The critical voltage (Vc) was defined as the voltage, after which Vmax was saturated. The insets showed the proposed model for the space charge effect. Qm, Qd, and Qf were the charges on metal electrodes, ionized dopants and free mobile carriers, respectively. + and indicated the polarity of the charges. Eac and Ein are the applied AC electric field and generated space charge field, respectively.
EL intensity was only determined by the carriers generated by the ionization of doped layers. In contrast, after the initiation of EL, it was a space charge modulated process [23]. The critical voltage depended on the AC voltage and frequency, as depicted in Fig. 7. In Fig. 7, Vac dependence of Vmax and D/ðV EL Þ was shown under various AC frequency. It can be seen that at lower frequency (1 Hz), the saturation behavior was well observed. Below Vc, Vmax was equal to AC voltage amplitude, i.e., EL intensity was maximum at AC amplitude. In this case, the space charge effect was insignificant. However, when Vac increased above Vc, Vmax was pinned at Vc. Although this behavior became difficult
L. Zhang et al. / Organic Electronics 15 (2014) 1815–1821
to distinguish under higher AC frequency (10 and 100 Hz), the increase of Vmax with AC amplitude gradually slowed down, implying that the critical voltage shifted to higher voltage outside of the measurement range. A higher frequency corresponded to a higher critical voltage and vice versa. In addition, at one AC voltage, EL phase approached to 90° with increasing frequency, which was consistent with Fig. 4. To further characterize the space charge effect, simulation is expected, which will be our next work. 4. Conclusions In conclusion, the AC responses of double-injection and double-insulated OLEDs were investigated and compared. For the double-injection OLEDs, the voltage-dependent transit time and frequency dependent carrier distribution were responsible for the observed phenomena. Although capacitive current was observed in the double-injection device, especially in the high-frequency region, the main contribution to the EL was attributed to the in-phase resistive current. On the other hand, for the double-insulated OLEDs, although they shared some characteristics with the double-injection ones, there were two differences between the two devices. One was the absence of resistive component and the other was the small EL phase difference (< 90 ) in the low-frequency region at high AC voltage. These phenomena were explained by considering the space charge effect, which caused the EL to last for a limited time interval. The phase difference analysis corresponded well with other results, but from a different standpoints with new information. Maximum EL intensity was found only when the EL phase difference was 90° in both devices, i.e., maximum EL intensity coincided with maximum AC voltage. The destroyed EL profiles due to the space charge effects were suggested to be responsible for the much lower EL intensity than that of the doubleinjection OLEDs and the EL intensity saturation at higher AC amplitude. To further improve device performance under AC driving conditions, on one hand, the charge generation ability should be enhanced with balanced carriers generation. On the other hand, the space charge effect should be carefully considered by optimizing the device structure and operating conditions (AC voltage and frequency). Acknowledgments This work was supported by a Grant-in-aid from the Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST) and the International Insti-
1821
tute for Carbon Neutral Energy Research (WPI-I2CNER) sponsored by the Ministry of Education, Culture, Sports, Science and Technology of Japan. References [1] S.R. Forrest, Nature 428 (2004) 911. [2] H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 492 (2012) 234. [3] W. Brütting, C. Adachi, Physics of Organic Semiconductors 2nd, Wiley-VCH, Berlin, 2012. [4] Y.Z. Wang, D.D. Gebler, L.B. Lin, J.W. Blatchford, S.W. Jessen, H.L. Wang, A.J. Epstein, Appl. Phys. Lett. 68 (1996) 894. [5] T. Tsutsui, S. Lee, K. Fujita, Appl. Phys. Lett. 85 (2004) 2382. [6] T. Yamao, Y. Shimizu, K. Terasaki, S. Hotta, Adv. Mater. 20 (2008) 4109. [7] J. Sung, Y.S. Choi, S.J. Kang, S.H. Cho, T. Lee, C. Park, Nano. Lett. 11 (2011) 966. [8] V. Wood, M.J. Panzer, D. Bozyigit, Y. Shirasaki, I. Rousseau, S. Geyer, M.G. Bawendi, V. Bulovic´, Nano. Lett. 11 (2011) 2927. [9] A. Perumal, M. Fröbel, S. Gorantla, T. Gemming, B. Lüssem, J. Eckert, K. Leo, Adv. Funct. Mater. 22 (2012) 210. [10] Z. Yang, B. Hu, F.E. Karasz, Macromolecules 28 (1995) 6151. [11] H.L. Wang, A.G. MacDiarmid, Y.Z. Wang, D.D. Gebler, A.J. Epstein, Synth. Met. 78 (1996) 33. [12] A.J. Pal, R. Österbacka, K.-M. Källman, H. Stubb, Appl. Phys. Lett. 70 (1997) 2022. [13] R. Österbacka, A.J. Pal, K.-M. Källman, H. Stubb, J. Appl. Phys. 83 (1998) 1748. [14] X.L. Xu, X.H. Chen, Y.B. Hou, Z. Xu, X.H. Yang, S.G. Yin, Z.J. Wang, X.R. Xu, S.P. Lau, B.K. Tay, Chem. Phys. Lett. 325 (2000) 420. [15] S.Y. Yang, L. Qian, F. Teng, Z. Xu, X.R. Xu, J. Appl. Phys. 97 (2005) 126101. [16] S. Lee, K. Fujita, T. Tsutsui, Jpn. J. Appl. Phys. 44 (2005) 6607. [17] J. Brückner, N. Christ, O. Bauder, C. Gärtner, M. Seyfried, F. Glückler, U. Lemmer, M. Gerken, Appl. Phys. Lett. 96 (2010) 041107. [18] Y. Chen, G.M. Smith, E. Loughman, Y. Li, W. Nie, D.L. Carroll, Org. Electron. 14 (2013) 8. [19] Y. Xia, Y. Chen, G.M. Smith, Y. Li, W. Huang, D.L. Carroll, Appl. Phys. Lett. 102 (2013) 253302. [20] A. Perumal, B. Lüssem, K. Leo, Appl. Phys. Lett. 100 (2012) 103307. [21] D. Bozyigit, V. Wood, Y. Shirasaki, V. Bulovic, J. Appl. Phys. 102 (2012) 113701. [22] M.A. Lampert, P. Mark, Current Injection in Solids, Academic Press, New York, 1970. [23] L. Zhang, D. Taguchi, T. Manaka, M. Iwamoto, J. Appl. Phys. 110 (2011) 033715. [24] L. Zhang, H. Nakanotani, C. Adachi, Appl. Phys. Lett. 103 (2013) 093301. [25] M. Fröbel, A. Perumal, T. Schwab, M.C. Gather, B. Lüssem, K. Leo, Org. Electron. 14 (2013) 809. [26] A. Perumal, B. Lüssem, K. Leo, Org. Electron. 13 (2012) 1589. [27] Y. Setoguchi, C. Adachi, J. Appl. Phys. 108 (2010) 064516. [28] D. Taguchi, S. Inoue, L. Zhang, J. Li, M. Weis, T. Manaka, M. Iwamoto, J. Phys. Chem. Lett. 1 (2010) 803. [29] D. Taguchi, L. Zhang, J. Li, M. Weis, T. Manaka, M. Iwamoto, J. Phys. Chem. C 114 (2010) 15136. [30] W. Brütting, S. Berleb, A.G. Muckl, Org. Electron. 2 (2001) 1. [31] S. Paul, P. Pal, A.J. Pal, G.B. Talapatra, J. Appl. Phys. 94 (2003) 4357. [32] S. Das, A. Chowdhury, A.J. Pal, Phys. Stat. Sol. (a) 185 (2001) 383. [33] D. Vaufrey, M.B. Khalifa, J. Tardy, C. Ghica, M.G. Blanchin, C. Sandu, J.A. Roger, Semicond. Sci. Technol. 18 (2003) 253. [34] H. Fujimoto, T. Miyayama, N. Sanada, C. Adachi, Org. Electron. 14 (2013) 2994.